Communication
doi.org/10.1002/chem.202101082
Chemistry—A European Journal
replacing the oxygen atom with C- as well as N-based
connectors (2p–s, Scheme 2d). Here, although N-allyl- and N-
Boc-indoline scaffolds 2p–r were isolated in moderate extents
(ee: 11–69%), the enantio-enriched dihydroindene acetic acid
2s was obtained in excellent stereochemical yield (ee=98%).
Then, the absolute configuration of compound 2a was
unambiguously determined to be R via single crystal X-ray
analysis of the corresponding bromo-amide 3a (Scheme 2e).
Additionally, the pivotal role of CO2 in generating the carboxylic
unit of targeted compounds 2 was determined via a labelled
13C-experiment. As a matter of fact, a full incorporation of
13CÀ carbon dioxide (99.8% labelling) was obtained in the final
compound 2a when 13CO2 was employed under optimal
reaction conditions (13C-2a, yield=65%, ee=98%, Scheme 2f).
In order to gain further insight into this reaction, a
mechanistic exploration was carried out in parallel by DFT
simulations (detailed methodology can be found in the
Supporting Information).[15] Several questions that are key to
the understanding of this reactivity are, at least: 1- what is the
structural model of enantiodiscrimination, 2- which is the active
catalyst and 3- how the fundamental role of the solvent can be
explained.[16]
Initially we assumed that the [(L)2Ni(H2O)Cl]Cl complex, with
a 2:1 L:Ni ratio, would dissociate delivering the LNi species I as
the active catalyst (see the NLE experiment).[17] We also assumed
that [Ni(0)] would be the oxidation state of the catalyst, after
reduction with the excess of zinc.[18] Catalytic cycles were
simulated via DFT calculations (for computational details see
the Supporting Information) for the PyOx (L3) and its imidazo-
line variant L11. Both ligands yielded similar reaction profiles.
First we explored a [Ni(0)]/[Ni(II)] catalytic cycle in which Zn
would only participate at the end, to restore the active [Ni(0)]
catalyst I (path A – red zone in Scheme 3). From this, a facile
oxidative addition of 1a occurs to form intermediate II. Then,
the stereodiscriminating addition of the NiÀ C bond to the
alkene must occur. The barriers associated to the formation of
the two diastereomers favor (by 2.5 kcal/mol) the formation of
intermediate III. We attempted to describe this step also on the
neutral complex II but we could only find the associated
transition state assuming prior loss of iodine. This is explained
through the need to open a coordinating vacant site such that
the double bond can be pre-activated for the addition step.
However, the insertion of CO2 onto intermediates III and III-
diast, to yield the final carboxylates IV and IV-diast, featured
high activation energies (26.3 and 36.7 kcal/mol, respectively)
rendering these paths unlikely.
We therefore decided to analyze whether a zinc-mediated
[Ni(II)]/[Ni(I)] reduction step along the reaction pathway could
facilitate the carboxylation event,[19] and indeed, the reaction
proceeds much more favorably if Ni is reduced right before CO2
insertion (path B – yellow block in Scheme 3). This reduction
step could also be occurring earlier in the mechanism, right
after the oxidative addition (path C – yellow block in Scheme 3).
Interestingly the latter alternative, which involves a rare Heck
step occurring at a Ni(I) species, also produces very competitive
barriers for the subsequent steps.
Scheme 3. Reaction mechanism explored via density functional theory. A
relative-free-energies computed at the PCM(DMF)-M06/Def2SVPP level of
theory at 1 atm and 298 K. See the Supporting Information for simulation
details.
To be able to determine which, of these two alternative
pathways (B or C), is more plausible and at what point along
the mechanism the reduction step is operating, we analyzed
our experimental results in detail. While doing so we realized
that one very common side product of this protocol is the
benzoic acid derivative (SP)[14] deriving by a direct carboxylation
of the arylÀ Ni intermediates. We therefore computed the
transition state for this carboxylation both at the [Ni(II)] and
[Ni(I)] complexes. We found that this step is very costly for the
[Ni(II)] complex (a computed barrier of about 30 kcal/mol) and
that it is feasible when acting on the [Ni(I)] species.[18] These
results therefore not only help explain the formation of this
byproduct but also strongly candidate the path B – yellow block
in Scheme 3 as the mechanism at work, involving an unusual
Heck step on a [Ni(I)] complex.[20]
In an attempt to unequivocally determine the nature of the
active catalyst, we performed a non-linear effect study[17] on the
model transformation 1a!2a, by varying the enantiopurity of
the chiral ligand L3. Interestingly, a perfect linear correlation
between ee(L3) and ee(2a) was observed (see Figure S1).
This finding led us to conclude that indeed the isolated
[(L11)2Ni(H2O)Cl]Cl species should be considered a pre-catalytic
unit, capable of delivering the active 1:1 active organometallic
species through an in situ ligand dissociation event.
Non-covalent interactions (NCI) analysis performed onto
transition states TSII–III and TSII–IIIdiast provides an interesting
picture of the stereodiscriminating mechanism where the tBu
Chem. Eur. J. 2021, 27, 1–7
4
© 2021 Wiley-VCH GmbH
��
These are not the final page numbers!